270 SCIENCE PROGRESS. 



tures, it was long supposed that the critical temperature for 

 a mixture was always between the r temperatures for the 

 components ; Pawlewski and Strauss found it to be ap- 

 proximately proportional to the composition. This law 

 does not take into account the difference between 4 and /p 

 and is, moreover, found to be confirmed only in a few- 

 instances. Mixtures of CO^ and CH3 CI and of COj and 

 H CI do not obey the law, the deviations being several 

 degrees, and higher for 4 than for /p. Sometimes the 

 critical temperatures for the mixtures are even partly 

 outside those for the constituents. Mixtures of N^O 

 (36°) and C2 Hg (32°) containing more than one-tenth of 

 Cj Hg have critical temperatures below 32° C. There is 

 no reason why more instances should not be discovered, 

 as so few accurate experiments have as yet been made in 

 this direction. The critical pressures were between those 

 for N.O and C^ Hg. The critical pressures for mixtures 

 of CO2 and CH3 CI are some of them higher than for 

 either of the substances. Evidently there is a great deal 

 of variety. Pictet uses the determination of the critical 

 constants (especially temperature) as a test for the purity 

 of substances. It is obvious from the foregoing that this can 

 only be a trustworthy method in very special cases where 

 the nature of the impurity is known beforehand. A much 

 better, but perhaps somewhat too delicate, test is the con- 

 stancy of the condensation pressure, though again this 

 method would fail altogether where the mixture was one 

 of maximum or minimum vapour-pressure. 



It is now necessary to point out the connection of the ex- 

 perimental results with the theory. In the above I have not 

 followed the historicalorder. Before anyaccurateexperiments 

 were made. Professor van der Waals published a theory for 

 mixtures without which the phenomena would not have been 

 easily unravelled. It is not inconceivable that most of 

 the results represented by our diagrams might have been 

 arrived at by a long and laborious set of experiments, but 

 practically it was the theory that led the way and even now 

 there is a good deal in the diagrams which depends on the 

 theory and is still waiting for experimental confirmation. 



